This article is the first installment in a series on e-discovery issues impacting inside counsel and areas that offer the greatest opportunity to reduce risks and save costs. This column looks at one of the most fundamental yet often ignored aspects of managing the production of electronically stored information: memorializing the process. 

Creating a well-documented record of the e-discovery process is critical. In determining whether parties have complied with production obligations in civil litigation, courts are guided by the touchstone of reasonableness. Even the simplest e-discovery project requires many decisions on various technical, legal and factual issues. And while every case is different, parties must generally be able to explain what they preserved, collected and produced—and why.

Of course, parties can’t show that what they did was reasonable, if they can’t show what they did. As outside litigation counsel, if an issue arises, I need to prove that what we did months or years earlier was reasonable, so I document as much as possible, generally including the following:

The litigation hold

When a new case comes in, one of the first questions is whether a litigation hold has been issued. Usually by the time a case gets to me, it should have been. This isn’t optional. Since the now-famous Zubulake opinions issued by Judge Scheindlin of the Southern District of New York in 2003 and 2004, most jurisdictions have come to follow a common-law standard that parties must issue a ligation hold—telling all key players (custodians) to preserve potentially relevant information—as soon as litigation becomes reasonably foreseeable. And, “oral” litigation holds are far less likely to pass muster, especially since that same judge held in a subsequent 2011 opinion that failure to issue a “writtenlitigation hold constitutes “gross negligence.” 

So make sure the litigation hold is in writing, and includes at minimum the date of issue, the recipients and the scope of preservation. Many parties include more in their holds, like reference to the triggering event that made litigation reasonably foreseeable, avenues for recipients to ask questions and advice of others who might be subject to the notice and/or an initial explanation of how the data will be collected.

Often companies exposed to repeat litigation will use a standardized hold, while others will need to create one customized for a particular litigation. In any event, be sure the folks in IT who are responsible for backing up and deleting data also get a copy and know how to implement it. In fact, it can be a good idea to provide a separate written notice order just for them. And don’t forget to re-issue the holds regularly (e.g. two or four times a year) if necessary to ensure continued compliance.

Retention policies, systems architecture, data maps and custodial interviews 

Another sea change from the Zubulake opinions came out of that court’s admonishment that counsel should become “fully familiar with her client’s document retention policies, as well as the client’s data retention architecture.” This, the court found, “will invariably involve speaking with information technology personnel, who can explain system-wide backup procedures and the actual (as opposed to theoretical) implementation of the firm’s recycle policy” and will involve “communicating with the ‘key players’ in the litigation, in order to understand how they stored information.” This implicates several areas that also should be memorialized. 

Document retention and destruction policies likely already have been put in writing and updated as necessary. Hopefully, the company complies with its policies. They should be shared with outside counsel.

Back when I was in law school, they did not teach “data retention architecture.” Moreover, every company I’ve ever worked for has a different information technology architecture, often including various flavors of document-management software, shared servers, proprietary databases and legacy systems. Thus, enlightened business organizations will often maintain “data maps” charting or describing in simple terms the IT systems, including where the various data types—from custom databases to e-mail and documents—are maintained. Many organizations, however, don’t keep data maps in the ordinary course of business, so outside counsel may have to create one to truly understand their client’s data retention architecture. 

Whether a data map exists, counsel in any sizable litigation should follow Judge Scheindlin’s advice and interview both IT personnel and key players. I keep two forms for this purpose: a system architecture form and a custodial interview form. The system architecture form either creates or updates the data map, and the custodial interview form documents interviews of the key players, memorializes where each keeps their data (including places like home computers and personal devices) and confirms the key players or custodians are aware of the obligation to preserve potentially relevant data. This can prove to a court that a collection was done reasonably and in good faith.

Conferences with opposing counsel

A later installment of this column will delve into the mechanics and benefits of a well-prepared Rule 26(f) conference (the conference that takes place between or among parties early in federal litigation). But whether conducted pursuant to rules, a court order or for other reasons, any representations to opposing counsel regarding the nature and scope of preservation or production should be memorialized. The drafters of the federal rules realized the importance of addressing e-discovery issues up front. If opposing parties who are made aware of limitations early on do not object in a timely fashion to what a producing party says it will do, courts will be more likely to invoke the doctrines of waiver and estoppel when those same parties come to complain of supposed production infirmities on the eve of trial.

Collecting, culling and sampling

Obviously, the process of collecting data from custodians, remote devices and company servers should be carefully documented. There should be chain-of-custody memoranda for every physical device or hard drive collected, as well as a thorough written explanation of the collection process. Usually, parties will over-collect, then preserve and cull the data.

Culling is the process of reducing a large tranche of collected data into one that is more reasonable and manageable. Without appropriate culling, the cost of processing and review can soar. Thus, it is not uncommon to “collect broad,” and cull the data by limiting date ranges, excluding certain file types, excluding email sent from certain domains (e.g., and further reducing through keyword searches. Plainly, in addition to memorializing what data they originally collected, parties should document the precise nature and justification for any culling.

On a related topic, keyword searches often will result in thousands or hundreds of thousands of hits. Assessing the quality of those hits is often done by sampling a manageable portion of the result to determine whether the search actually achieved usable data. Courts often are persuaded by results of well-thought-out sampling, but the issue may only arise months or years after the searches were first performed. Thus, records of any sampling should be carefully maintained.

Quality control

Many document reviews involve roomfuls of contract lawyers reviewing documents on computer terminals for months on end. But how can a party defend itself against sanctions for not producing the document that was—in good faith—simply missed?

The best protection is to be able to show that the review was reasonably conducted, and quality control usually plays a big role. Not only must the reviewers be well trained, but a good quality control (QC) often involves random sampling of the results of their review (e.g., by having a more senior attorney re-review every 50th or 100th document to confirm the initial analysis was correct). Thus, be sure to keep all materials provided to and used to train the document reviewers, as well as the results of QC checks for each reviewer.


The collection, processing, review and production of electronically stored information can be the most expensive aspect of litigation. This expense will multiply if sanctions are brought for failure to properly preserve or produce evidence. So be careful to memorialize the process. It can be your best insurance policy.

In future installments of this series, we will address self-collection, minimizing processing and review costs, managing a review team, agreements with opposing counsel and the problem of international e-discovery.